package PPI::Find; =pod =head1 NAME PPI::Find - Object version of the Element->find method =head1 SYNOPSIS # Create the Find object my $Find = PPI::Find->new( \&wanted ); # Return all matching Elements as a list my @found = $Find->in( $Document ); # Can we find any matching Elements if ( $Find->any_matches($Document) ) { print "Found at least one matching Element"; } # Use the object as an iterator $Find->start($Document) or die "Failed to execute search"; while ( my $token = $Find->match ) { ... } =head1 DESCRIPTION PPI::Find is the primary PDOM searching class in the core PPI package. =head2 History It became quite obvious during the development of PPI that many of the modules that would be built on top of it were going to need large numbers of saved, storable or easily creatable search objects that could be reused a number of times. Although the internal ->find method provides a basic ability to search, it is by no means thorough. PPI::Find attempts to resolve this problem. =head2 Structure and Style PPI::Find provides a similar API to the popular L module for file searching, but without the ability to assemble queries. The implementation of a separate PPI::Find::Rule sub-class that does provide this ability is left as an exercise for the reader. =head2 The &wanted function At the core of each PPI::Find object is a "wanted" function that is passed a number of arguments and returns a value which controls the flow of the search. As the search executes, each Element will be passed to the wanted function in depth-first order. It will be provided with two arguments. The current Element to test as $_[0], and the top-level Element of the search as $_[1]. The &wanted function is expected to return 1 (positive) if the Element matches the condition, 0 (false) if it does not, and undef (undefined) if the condition does not match, and the Find search should not descend to any of the current Element's children. Errors should be reported from the &wanted function via die, which will be caught by the Find object and returned as an error. =head1 METHODS =cut use strict; use Params::Util qw{_INSTANCE}; use vars qw{$VERSION}; BEGIN { $VERSION = '1.236'; } ##################################################################### # Constructor =pod =head2 new &wanted The C constructor takes a single argument of the &wanted function, as described above and creates a new search. Returns a new PPI::Find object, or C if not passed a CODE reference. =cut sub new { my $class = ref $_[0] ? ref shift : shift; my $wanted = ref $_[0] eq 'CODE' ? shift : return undef; # Create the object my $self = bless { wanted => $wanted, }, $class; $self; } =pod =head2 clone The C method creates another instance of the same Find object. The cloning is done safely, so if your existing Find object is in the middle of an iteration, the cloned Find object will not also be in the iteration and can be safely used independently. Returns a duplicate PPI::Find object. =cut sub clone { my $self = ref $_[0] ? shift : die "->clone can only be called as an object method"; my $class = ref $self; # Create the object my $clone = bless { wanted => $self->{wanted}, }, $class; $clone; } #################################################################### # Search Execution Methods =pod =head2 in $Document [, array_ref => 1 ] The C method starts and completes a full run of the search. It takes as argument a single L object which will serve as the top of the search process. Returns a list of PPI::Element objects that match the condition described by the &wanted function, or the null list on error. You should check the ->errstr method for any errors if you are returned the null list, which may also mean simply that no Elements were found that matched the condition. Because of this need to explicitly check for errors, an alternative return value mechanism is provide. If you pass the C<< array_ref => 1 >> parameter to the method, it will return the list of matched Elements as a reference to an ARRAY. The method will return false if no elements were matched, or C on error. The ->errstr method can still be used to get the error message as normal. =cut sub in { my $self = shift; my $Element = shift; my %params = @_; delete $self->{errstr}; # Are we already acting as an iterator if ( $self->{in} ) { return $self->_error('->in called while another search is in progress', %params); } # Get the root element for the search unless ( _INSTANCE($Element, 'PPI::Element') ) { return $self->_error('->in was not passed a PPI::Element object', %params); } # Prepare the search $self->{in} = $Element; $self->{matches} = []; # Execute the search if ( !eval { $self->_execute; 1 } ) { my $errstr = $@; $errstr =~ s/\s+at\s+line\s+.+$//; return $self->_error("Error while searching: $errstr", %params); } # Clean up and return delete $self->{in}; if ( $params{array_ref} ) { if ( @{$self->{matches}} ) { return delete $self->{matches}; } delete $self->{matches}; return ''; } # Return as a list my $matches = delete $self->{matches}; @$matches; } =pod =head2 start $Element The C method lets the Find object act as an iterator. The method is passed the parent PPI::Element object as for the C method, but does not accept any parameters. To simplify error handling, the entire search is done at once, with the results cached and provided as-requested. Returns true if the search completes, and false on error. =cut sub start { my $self = shift; my $Element = shift; delete $self->{errstr}; # Are we already acting as an iterator if ( $self->{in} ) { return $self->_error('->in called while another search is in progress'); } # Get the root element for the search unless ( _INSTANCE($Element, 'PPI::Element') ) { return $self->_error('->in was not passed a PPI::Element object'); } # Prepare the search $self->{in} = $Element; $self->{matches} = []; # Execute the search if ( !eval { $self->_execute; 1 } ) { my $errstr = $@; $errstr =~ s/\s+at\s+line\s+.+$//; $self->_error("Error while searching: $errstr"); return undef; } 1; } =pod =head2 match The C method returns the next matching Element in the iteration. Returns a PPI::Element object, or C if there are no remaining Elements to be returned. =cut sub match { my $self = shift; return undef unless $self->{matches}; # Fetch and return the next match my $match = shift @{$self->{matches}}; return $match if $match; $self->finish; undef; } =pod =head2 finish The C method provides a mechanism to end iteration if you wish to stop the iteration prematurely. It resets the Find object and allows it to be safely reused. A Find object will be automatically finished when C returns false. This means you should only need to call C when you stop iterating early. You may safely call this method even when not iterating and it will return without failure. Always returns true =cut sub finish { my $self = shift; delete $self->{in}; delete $self->{matches}; delete $self->{errstr}; 1; } ##################################################################### # Support Methods and Error Handling sub _execute { my $self = shift; my $wanted = $self->{wanted}; my @queue = ( $self->{in} ); # Pull entries off the queue and hand them off to the wanted function while ( my $Element = shift @queue ) { my $rv = &$wanted( $Element, $self->{in} ); # Add to the matches if returns true push @{$self->{matches}}, $Element if $rv; # Continue and don't descend if it returned undef # or if it doesn't have children next unless defined $rv; next unless $Element->isa('PPI::Node'); # Add the children to the head of the queue if ( $Element->isa('PPI::Structure') ) { unshift @queue, $Element->finish if $Element->finish; unshift @queue, $Element->children; unshift @queue, $Element->start if $Element->start; } else { unshift @queue, $Element->children; } } 1; } =pod =head2 errstr The C method returns the error messages when a given PPI::Find object fails any action. Returns a string, or C if there is no error. =cut sub errstr { shift->{errstr}; } sub _error { my $self = shift; $self->{errstr} = shift; my %params = @_; $params{array_ref} ? undef : (); } 1; =pod =head1 TO DO - Implement the L class =head1 SUPPORT See the L in the main module. =head1 AUTHOR Adam Kennedy Eadamk@cpan.orgE =head1 COPYRIGHT Copyright 2001 - 2011 Adam Kennedy. This program is free software; you can redistribute it and/or modify it under the same terms as Perl itself. The full text of the license can be found in the LICENSE file included with this module. =cut